COE308 Hm 1 Sol

Embed Size (px)

Citation preview

  • 7/29/2019 COE308 Hm 1 Sol

    1/3

    King Fahd University of Petroleum and Minerals

    Department of Computer Engineering

    COMPUTER ARCHITECTURE COE 308

    Homework 1

    Problems Grading

    12

    3

    4

    5

    TOTAL

  • 7/29/2019 COE308 Hm 1 Sol

    2/3

    QUESTION 1: Trends in Computer Technology

    Shortly summarize the growth rate in computer performance every year and provide justification.

    Refer to annual: (1) microprocessor clock rate, (2) chip density (transistor), (3) DRAM and Disk

    access times, (4) DRAM and Disk memory capacity, and (5) usage of address bits. Determine the

    new trends in computer architecture with respect to Internet, networking and multimedia

    requirements.

    Solution:

    The rate of growth is: (1) 50% for the peak MIPS speed, (2) 66% for the number of transistors (chipdensity), (3) 5 times a year for the memory and disk capacity, and (4) 7% for the memory access

    time, (5) usage of address bits is increased by 20% per year. The CPU architecture is shifting from

    the traditional computing computer to the multimedia and internet computers. This has deep

    consequences on the architecture. The unpredictable hierarchical memory is not likely to be very

    useful because of the need for predictable real-time processing for multimedia. I/O streaming

    between memory and the display is another important aspect for future computers. The new trends in

    computer architecture are due to the Internet, networking, and multimedia requirements involving

    extensive use of computer graphics, SIMD CPU extension with the use of narrow data (bytes), I/O

    streaming to shorten the path to display, real-time aspects and worst-case guaranteed performance to

    support QoS, support to fine-grain and coarse-grain parallelism for advanced multimedia algorithms,

    high code locality, very high network bandwidth, etc.

    QUESTION 2: Processor-Memory Performance Gap

    1. Since year 2000 the clock frequency of Intel IA32 processors is growing at a rate of

    approximately 1.54 per year. In 2005, the reference clock rate is 3 GHz. What is the expected

    clock frequency CF of IA32 processor in year 2010.

    2. The access time of some DRAM memory is shrinking by a factor of 6% per year. It is estimated

    to be 27 ns in year 2005. What is the expected access time AT of the same DRAM in year 2010.

    3. Denote by N(k) the ratio of the DRAM access time over the processor clock time, where k is the

    year. Evaluate N for k=2005 and k=2010 and comment on growth rate of N.

    Solution:

    1. The expected processor clock frequency in year 2010 is CF = 3 * (1.54)^5 = 25.986 GHz2. The expected access time of DRAM in year 2010 is AT = 27 * (0.94)^5 = 19.82 ns

    3. The expected value of N in year 2005 is N= AT / (1/CF) = AT * CF = 27 * 3 = 81. Therefore, in

    year 2010 we have N = 27 * (0.94)^5 * 3 * (1.54)^5 = 27 * 3* (0.94*1.54)^5 = 514.92. The

    gap between processor speed and DRAM access time is growing which is an indicator of the

    need of new techniques to improve performance of cache memory.

    QUESTION 3: MIPS Assembly Language Programming

    Consider the following fragment of C code:

    for (i=0; i

  • 7/29/2019 COE308 Hm 1 Sol

    3/3

    add $s0, $s0, 1 ; Add 1 to N to exit when s0=0Loop: sub $s0, $s0, 1 ; decrement N

    sub $t0, $t0, 4 ; increment the index i by 4

    add $t2, $t1, 1 ; evaluate pointer i for V[]

    add $t4, $t3, 4 ; evaluate pointer i for W[]

    Lw $t5, 0($t2) ; load array element V[i]

    Lw $t6, 0($t4) ; load array element W[i]

    Add $t5, $t5, $t6 ; compute V[i]= V[i]+ W[i]

    Sw $t5, 0($t2) ; store the value of V[i]+ W[i]

    bneq $s0, $zero, loop ; Loop if N is not zero

    The number of instructions are executed during the running of this code is 9x100+1=901

    The number of memory data references made during execution is 3*100=300

    QUESTION 4: Internet search (No solution will be provided)

    Search on the Internet for current technology on the following issues:

    1. The RISC and CISC architectures, comparing them, and giving examples of processors from

    each class.2. The most recent PC-based DRAM technology, and determine its most important functional

    features.

    3. The most recent PC-based INTEL microprocessor, summarize its architecture, determine its

    most important functional features, and present the type of services that are targeted by its

    design.

    For each of the above the student is expected to summarize his finding and attach a copy of the

    referenced papers.